Function pad

pywt/_dwt.py:404–516

Extend a 1D signal using a given boundary mode. This function operates like :func:`numpy.pad` but supports all signal extension modes that can be used by PyWavelets discrete wavelet transforms. Parameters ---------- x : ndarray The array to pad pad_widths : {sequenc

(x, pad_widths, mode)

def pad(x, pad_widths, mode):
    """Extend a 1D signal using a given boundary mode.

    This function operates like :func:`numpy.pad` but supports all signal
    extension modes that can be used by PyWavelets discrete wavelet transforms.

    Parameters
    ----------
    x : ndarray
        The array to pad
    pad_widths : {sequence, array_like, int}
        Number of values padded to the edges of each axis.
        ``((before_1, after_1), … (before_N, after_N))`` unique pad widths for
        each axis. ``((before, after),)`` yields same before and after pad for
        each axis. ``(pad,)`` or int is a shortcut for
        ``before = after = pad width`` for all axes.
    mode : str, optional
        Signal extension mode, see :ref:`Modes <ref-modes>`.

    Returns
    -------
    pad : ndarray
        Padded array of rank equal to array with shape increased according to
        ``pad_widths``.

    Notes
    -----
    The performance of padding in dimensions > 1 may be substantially slower
    for modes ``'smooth'`` and ``'antisymmetric'`` as these modes are not
    supported efficiently by the underlying :func:`numpy.pad` function.

    Note that the behavior of the ``'constant'`` mode here follows the
    PyWavelets convention which is different from NumPy (it is equivalent to
    ``mode='edge'`` in :func:`numpy.pad`).
    """
    x = np.asanyarray(x)

    # process pad_widths exactly as in numpy.pad
    pad_widths = np.array(pad_widths)
    pad_widths = np.round(pad_widths).astype(np.intp, copy=False)
    if pad_widths.min() < 0:
        raise ValueError("pad_widths must be > 0")
    pad_widths = np.broadcast_to(pad_widths, (x.ndim, 2)).tolist()

    if mode in ['symmetric', 'reflect']:
        xp = np.pad(x, pad_widths, mode=mode)
    elif mode in ['periodic', 'periodization']:
        if mode == 'periodization':
            # Promote odd-sized dimensions to even length by duplicating the
            # last value.
            edge_pad_widths = [(0, x.shape[ax] % 2)
                               for ax in range(x.ndim)]
            x = np.pad(x, edge_pad_widths, mode='edge')
        xp = np.pad(x, pad_widths, mode='wrap')
    elif mode == 'zero':
        xp = np.pad(x, pad_widths, mode='constant', constant_values=0)
    elif mode == 'constant':
        xp = np.pad(x, pad_widths, mode='edge')